Germany's CO₂ Pipeline Puzzle Has a Surprising Answer: 500 Kilometers Is Enough

Germany has promised to stop adding carbon to the atmosphere by 2045, but it has a stubborn problem: a country cannot electrify its way out of making cement. The limestone chemistry is what it is; heating the kiln with green power doesn't change that. So a growing consensus has formed, from the European Net-Zero Industrial Act to Germany's own Carbon Management Law, that some factories will need to capture their carbon dioxide and bury it or ship it somewhere it can be stored. The question was never really whether to build the pipelines. The question is which pipelines to build first.

That question turns out to be surprisingly delicate. Capture plants are expensive, and nobody wants to bolt a carbon-capture retrofit onto a cement plant if there's no pipe to carry the gas away. But nobody wants to pay for a pipe that runs to a region where the factories never hook up. It's a chicken-and-egg problem with 50-year lifetimes on both sides. A new study from researchers at TU Berlin and Fraunhofer ISI tries to break that deadlock by combining two very different ways of thinking about infrastructure, and it lands on a refreshingly practical answer: the exact route matters far less than simply having a connection that works. Once the biggest industrial regions are wired to a place that can store carbon, almost all of the economic benefit arrives — with as little as 500 kilometers of pipe (Seibold et al., 2026).

The Science

The researchers, Toni Seibold, Luna Lütz, and Tom Brown, faced a classic problem in energy modeling. On one side, you have detailed infrastructure studies that map out precise corridors, account for compressor stations, and ponder whether to move the carbon dioxide as a gas, a dense liquid, or by ship. These are wonderfully specific, but they take the wider energy system as a given — they ask what the pipes should look like in a world that is already decarbonized, usually in the far-off target year of 2050)Skip.

On the other side, you have large integrated energy system models like PyPSA-DE, which represent the whole economy — electricity, heating, transport, industry, agriculture — and optimize all of it together to minimize annual system cost. These models capture the feedbacks beautifully but tend to treat pipelines as continuous, smoothly expanding capacity rather than as the lumpy, discrete, you-build-it-or-you-don't decisions that real planners face)Skip.

Seibold and colleagues built a bridge between the two. First they used a graph-theoretic algorithm to generate candidate pipeline topologies. Think of Germany as a spatial graph: each node is a clustered region with a certain "capturable" carbon potential from industrial point sources and waste incineration, and each edge is a potential pipeline corridor with a physical length. The algorithm grows a network outward from a chosen sink — the place where carbon is stored — using a weighted cost that balances distance against how much carbon a region can potentially supply. The weighting parameter $\alpha$ controls the trade-off: set it to zero and you get the shortest, most compact network regardless of where the carbon is; raise it and the corridors bend toward high-potential industrial regions.

From this single machinery, the team generated 60 distinct candidate topologies. Some were single-sink trees rooted in one storage location; others grew multiple strands toward several sinks at once, in a forest of sink-oriented branches. Some variants deliberately perturbed the assumed carbon potential of up to three connected regions, reflecting genuine uncertainty about whether individual plants will actually retrofit capture, which can force the heuristic onto alternative routes. The topologies differed in total length — with budgets ranging from 500 to 1500 kilometers — in how accessible the sinks were, and in which sources got prioritized.

Each candidate topology was then handed to PyPSA-DE, a sector-coupled model that represents every major industrial process from cement and steel to methanol and ammonia at the level of individual plants, including their retirement dates.

Here's the crucial part: the energy system model then gets to decide, endogenously, what actually flows through those pipes. It decides how much carbon to capture, whether to bury it or use it to synthesize methanol, Fischer-Tropsch fuels, or methane, and how much to compress and pump. One small concession to physics: the model represents pipelines in dense phase onlySolve upstream. It applies a routing factor of 1.25 to centroid distances to account for pipes that can't run in a perfectly straight line.

This hybrid approach gives you something neither method could deliver alone: discrete, interpretable, lumpy pipelines — the first few that get built — embedded in a full economy that responds to them.

What They Found

The headline number is striking. Sinking a domestic carbon transport network into the model reduces German consumer costs by around 22 billion euros per year relative to building no pipelines at all (Seibold et al., 2026). That's a figure large enough to dwarf the cost of the pipes themselves, because without transport, the expensive part of the hard-to-abate problem has no way out.

But the more interesting findings come from looking at what survives across all 60 topologies. The team tallied how often each corridor was selected across different sink assumptions, weighting parameters, and perturbations. The result, shown in Figure 4 of the paper (

), is a map dominated by dark blue in north-western Germany. A handful of corridors emerge again and again no matter what assumptions are thrown at the algorithm: the routes connecting the industrial heartland of North Rhine-Westphalia toward the Dutch coast, where access to North Sea sequestration begins.

The utilization picture deepens this. Figure 7

overlays the selection frequency with the actual average carbon throughput those corridors carry in 2035, and the two align tightly. The corridors that get chosen are the ones that get used, and they're used heavily. The carbon flowing through them comes overwhelmingly from industrial point sources: cement clinker production, process emissions, steelmaking, and biomass-based carbon dioxide removal that counts as negative emissions. The contribution from backup power generation — natural gas plants that might run at moments of high demand — is comparatively small, barely registering in the throughput. That's worth holding onto: the first pipelines in Germany are industrial arteries, not power-sector convenience.

The single most policy-relevant result concerns how much pipeline you actually need. Comparing a 1500-kilometer network against a modest 500-kilometer one, the researchers found that the short network captures most of the economic benefit — provided it connects North Rhine-Westphalia to the Netherlands. This is the corridor that matters. Early access to Dutch sink infrastructure delivers outsized system value, because it links the densest cluster of industrial emitters to a storage route without waiting for a purely German offshore solution to materialize. And there's a second, more subtle benefit: building the short network now doesn't lock the country into a particular long-term topology. It leaves options open. You get most of the value today while avoiding the risk of entrenching a route that a more mature system might not want, a phenomenon economists call "infrastructure lock-in" (Seibold et al., 2026).

The paper's interpretation crystallizes into a single claim: once the major industrial source regions and sink access points are connected, the availability of carbon transport infrastructure matters far more than the exact topology. The 2035 system is remarkably insensitive to whether the pipes snake one way or another, as long as the big sources can reach a sink.

Why This Changes Things

This finding defuses what could have been a very expensive argument. If the economic value of a carbon network depended crucially on getting the precise geometry right, planners would face a paralyzing optimization problem under deep uncertainty — you'd want to model every route perfectly before committing a single euro. The study suggests the opposite: the decision space is forgiving. Get the big regions connected and the rest is detail.

There's a broader lesson hiding here about how to reconcile two research cultures. Infrastructure studies and energy system models have long talked past each other. Detailed bottom-up work gives you beautiful corridor maps but can't tell you how the wider energy system will respond, while big system models smooth away the lumpy reality of "you build this pipe or you don't." By generating discrete topologies and then letting the model react to them, this paper shows the two can be married — and that the marriage is what produces genuinely decision-relevant insight about sequencing, not just end states.

The 500-kilometer result also reframes the conversation around German carbon infrastructure from a question of how vast a build-out is needed to a question of which single spine matters most. It's a genuinely anti-grandiose conclusion. The instinct in infrastructure planning is often to think big — build the full backbone, get ahead of demand. This work suggests a leaner, more surgical first step: a focused connection from the Ruhr to the Dutch coast. That's a very different kind of advice than the usual "we need thousands of kilometers of pipe."

One caution worth emphasizing: the 500-kilometer network captures most of the 2035 economic benefit. The study is explicit that its transport focus is a 2035 target, which means it's designed to answer the start-up question, not the full climate-neutrality question. The model's sequestration sensitivity — lowering the assumed availability of the Dutch sinks or of a German North Sea route — shifts which corridors look attractive. In the sequestration-sensitivity scenario, shown in Figure 23 (

), and the German North Sea availability scenario in Figure 26

, the optimal choices bend accordingly. When a purely German offshore solution is assumed available, the corridors rearrange to feed it; when it isn't, the Dutch connection becomes even more central.

What's Next

The study is honest about its limits. Direct air capture is deliberately absent from the 2035 horizon, on the reasonable grounds that it won't deploy at scale in Europe that soon — but that means the long-term picture, where removal technologies loom larger, isn't fully resolved here. The model also treats industrial production levels as exogenous, following the German Environment Agency's "Further measures" scenario, in which industrial final energy demand declines from 614 TWh in 2025 to 576 TWh in 2035. That's a defensible assumption, but it means the results don't speak to worlds where Germany's industrial structure looks different.

The biggest open question is whether the short, surgical network remains the right answer as the system matures. A 500-kilometer Dutch spine is a superb first step, but the full climate-neutrality transition will demand more. Future work will need to extend this hybrid method beyond a single target year — to let the early decisions and the later build-out speak to each other across time, which is precisely where lock-in risk lives. The authors' framework, which generates candidate topologies and evaluates them inside a full sector-coupled model, is well positioned to do exactly that.

There's also a quieter significance here that extends well beyond Germany. Europe is about to spend heavily on carbon transport, with flagship projects like Northern Lights, Porthos, and the Delta Rhine Corridor all promising to move captured CO₂ to North Sea storage. Every one of those projects faces the same co-dependency that this paper tackles: capture investments need transport, and transport needs confidence in capture. The message — that connecting the biggest sources to a usable sink captures nearly all the value, that the exact route can be refined later, that early international access to storage infrastructure is disproportionately valuable — is not a German lesson. It's a general one for every industrial region staring at its residual emissions and wondering where, exactly, to lay a pipe.

The low-carbon transition is often described as a problem of physics and economics. This paper is a reminder that it's also a problem of sequencing — of deciding what to build first when every investment depends on anotherSkip. And the good news is that, at least for the first step, the sequencing question turns out to be easier than feared: find your biggest sources, find your sink, connect them, and let the details take care of themselves.